Activators of Cylindrical Proteases as Antimicrobials: Identification and Development of Small Molecule Activators of ClpP Protease

Chemistry & Biology

Article

Activators of Cylindrical Proteases asAntimicrobials: Identification and Development ofSmall Molecule Activators of ClpP ProteaseElisa Leung,1 Alessandro Datti,2,3 Michele Cossette,4 Jordan Goodreid,4 Shannon E. McCaw,5 Michelle Mah,5

Alina Nakhamchik,5 Koji Ogata,6,11 Majida El Bakkouri,1 Yi-Qiang Cheng,7 Shoshana J. Wodak,5,6,8 Bryan T. Eger,1,5,9,10

Emil F. Pai,1,5,9,10 Jun Liu,5 Scott Gray-Owen,5 Robert A. Batey,4 and Walid A. Houry1,*1Department of Biochemistry, University of Toronto, Toronto, ON M5S 1A8, Canada2Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON M5G 1X5, Canada3Department of Experimental Medicine and Biochemical Sciences, University of Perugia, Perugia 06126, Italy4Department of Chemistry5Department of Molecular GeneticsUniversity of Toronto, Toronto, ON M5S 1A8, Canada6Centre for Computational Biology, Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada7Department of Biological Sciences and Department of Chemistry and Biochemistry, University of Wisconsin–Milwaukee, Milwaukee,

WI 53211, USA8Molecular Structure and Function Program, Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada9Department of Medical Biophysics, University of Toronto, Toronto, ON M5S 1A8, Canada10Division of Cancer Genomics and Proteomics, Ontario Cancer Institute/Princess Margaret Hospital, Campbell Family Institute for Cancer

Research, Toronto, ON M5G 1L7, Canada11Present address: RIKEN, Innovation Center, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan

*Correspondence: [email protected]

DOI 10.1016/j.chembiol.2011.07.023

SUMMARY

ClpP is a cylindrical serine protease whose ability todegrade proteins is regulated by the unfoldase ATP-dependent chaperones. ClpP on its own can onlydegrade small peptides. Here, we used ClpP asa target in a high-throughput screen for compounds,which activate the protease and allow it to degradelarger proteins, hence, abolishing the specificityarising from the ATP-dependent chaperones. Ourscreen resulted in five distinct compounds, which wedesignate as Activators of Self-CompartmentalizingProteases 1 to 5 (ACP1 to 5). The compounds arefound to stabilize the ClpP double-ring structure.The ACP1 chemical structure was considered tohave drug-like characteristics and was further opti-mized to give analogs with bactericidal activity.Hence, the ACPs represent classes of compoundsthat can activate ClpP and that can be developedas potential novel antibiotics.

INTRODUCTION

In recent years, there has been an alarming trend of increased

bacterial infections caused by strains resistant to most known

antibiotics. As a result, diseases that were thought to be

controlled by currently available drugs are re-emerging not

only in developing countries but also in industrialized nations,

especially in clinical settings such as hospitals. Therefore, there

Chemistry & Biology 18, 1167–117

is an urgent need for the development of new types of antibiotics

that can be used to effectively treat multidrug-resistant bacteria.

The development of new drugs with novel mechanisms of action

is clearly needed to avert an impending crisis.

Recently, a novel antibacterial target was identified when

Brotz-Oesterhelt et al. (2005) discovered that the caseinolytic

protease P, ClpP, is activated by acyldepsipeptides, ADEPs,

a class of compounds that were first reported to have antibiotic

properties in 1985 (Michel and Kastner, 1985). The ADEPs were

later chemically optimized to address issues related to potency

and aqueous solubility (Hinzen et al., 2006). The protein target

of the ADEPs, ClpP protease, is a tetradecameric serine

protease comprised of two stacked heptameric rings, which, in

Escherichia coli, can form complexes with the AAA+ ATPase

chaperones ClpX or ClpA (Katayama et al., 1988; Maurizi and

Xia, 2004). ClpX and ClpA are hexameric chaperones that bind

on one or both ends of the ClpP protease. The chaperones

bind to target proteins, unfold them, and then thread them into

the ClpP proteolytic chamber through axial pores lined by axial

loops for degradation. These activities require ATP (Gottesman

et al., 1997). In the absence of the ATPase components, ClpP

alone can efficiently degrade small peptides of up to about 30

amino acids (Gottesman et al., 1997) and can also degrade

unstructured proteins albeit with much lower efficiency when

compared with ClpXP or ClpAP (Jennings et al., 2008; Bewley

et al., 2009). ADEPs enhance the efficiency of ClpP-dependent

degradation of unstructured proteins by opening up the ClpP

axial pores (Lee et al., 2010; Li et al., 2010).

The aforementioned studies demonstrated that ClpP is an

attractive target for developing new antibiotics with a novel

mode of action. In this study, we have identified, by using

a high-throughput screening approach, several structurally

8, September 23, 2011 ª2011 Elsevier Ltd All rights reserved 1167

mailto:[email protected]

http://dx.doi.org/10.1016/j.chembiol.2011.07.023

Figure 1. Results of the High-Throughput

Compound Screen for Activators of ClpP

(A) About 60,000 compounds were screened to find acti-

vators of E. coli ClpP. B-score values were calculated for

each compound from the increase in fluorescence inten-

sity after a 6 hour incubation of casein-FITC with ClpP and

compound. Compounds confirmed as hits, designated

ACP1 through ACP5, are indicated by the arrows (see also

Figure S1). Other compounds with high B-scores were

false hits.

(B) Shown are chemical structures of ACP1-ACP5 as

well as of ACP1a and ACP1b. ACP1 is N-1-[2-(phenylthio)

ethyl]-2-methyl-2-{[5-(trifluoromethyl)-2-pyridyl]sulfonyl}pro-

panamide, ACP2 is 3-(tertbutoxy)-2-{[2-[(5-(tertbutoxy)-2-

{[(9-H-9-fluorenylmethoxy)carbonyl]amino}-5-oxopentanoyl)

amino]-3-(tertbutylsulfanyl)propanoyl]amino}butanoic acid,

ACP3 is [4-(7-chloroquinolin-4-yl)piperazino](cyclohexyl)

methanone, ACP4 is ethyl 2-(2,2-dichlorovinyl)-4-hydroxy-

4-(3-nitrophenyl)-6-oxocyclohexanecarboxylate, and

ACP5 is ethyl 4-(4-bromophenyl)-2-(2,2-dichlorovinyl)-4-

hydroxy-6-oxocyclohexanecarboxylate.

Chemistry & Biology

Activators of Self-Compartmentalizing Proteases

diverse non-ADEP compounds that activate the ClpP proteolytic

core to degrade protein substrates in the absence of the ClpX

or ClpA ATPases. These newly identified compounds have

been termed Activators of Self-Compartmentalizing Proteases

(ACP). Their chemical structures differ significantly from the

structures of the previously identified ADEPs. The chemical opti-

mization of ACP1 resulted in analogs having improved bioactivity

and bactericidal effects. Hence, our study provides the basis for

the development of novel antibiotics based on the activation and

dysregulation of ClpP activity using different structural scaffolds.

RESULTS

High-Throughput Screen for ClpP ActivatorsTo identify compounds that activate ClpP, we developed a high-

throughput screening assay with a fluorescence-based readout.

1168 Chemistry & Biology 18, 1167–1178, September 23, 2011 ª2011 Elsevier Ltd All

The assay employed fluorescein isothiocya-

nate-labeled casein (casein-FITC) as the

proteolytic target of the E. coli ClpP protease.

When casein-FITC is intact, FITC fluorescence

is quenched; protease-catalyzed hydrolysis of

casein-FITC relieves this quenching, yielding

highly fluorescent dye-labeled peptides. The

principle of the screen was to select for

compounds that result in increased fluores-

cence upon incubation of casein-FITC with

ClpP. Preliminary tests performed in the pres-

ence and absence of the unfoldase chaperone

ClpA revealed a 5-fold dynamic range after

30 min incubation and intra- and interassay

variability, expressed as coefficient of variation,

of 2% and 5%, respectively. In light of ClpP

stability over several hours at 37�C, reactionswere typically monitored every 15 min for 6 hr

to rule out time-dependent effects.

With the initial intent of exploring drug-reposi-

tioning opportunities, we employed 4500 chem-

icals composed of biologically and pharmaco-

logically active entities, of which approximately 45% were

marketed drugs or drug candidates evaluated in clinical trial

stages. Notably, none of these compounds were active at

a concentration of 10 mM, suggesting the necessity to signifi-

cantly broaden chemical diversity to explore ClpP druggability

and the likelihood to activate ClpP using small molecules.

Thus, we expanded the screening campaign to include addi-

tional �60,000 highly diverse, drug-like chemicals. This under-

taking was carried out using a final compound concentration of

20 mM. Results were normalized and corrected for systematic

errors using the B-scoremethod (Brideau et al., 2003) (Figure 1A)

and positive hits were defined as the compounds whose signals

were at least three standard deviations (99.73% confidence

interval) from the mean of the general sample population. An

excellent quality of the screening setup was shown by the

dimensionless parameters Z0- and Z-factors (Zhang et al.,

rights reserved

A

B

0

0.2

0.4

0.6

0.8

1.0

1.2

0 20 40 60 80 100 120

Rel

ativ

e Fl

uore

scen

ce C

hang

e

ACP1

ACP2ACP3

ACP4ACP5

ClpAP

Compound (μM)

after 6 hours

Compound RD25 SDACP1 0.53 0.04ACP2 0.20 0.01ACP3 0.10 0.04ACP4 0.37 0.05ACP5 0.39 0.04

Figure 2. Relative Degradation Index

(A) Shown is the effect of compound concentration on casein-FITC degrada-

tion by ClpP after 6 hour incubation. Data are the average of three repeats.

Error bars represent standard deviations.

(B) Comparison of the RD25 values for the ACP compounds. Data shown

represent the average of three repeats. SD is standard deviation.

Chemistry & Biology


1999), which were consistently in the 0.7 range throughout the

entire screening campaign, thereby, indicating an effective

combination of dynamic range, variability, and hit rate.

This chemical screen led to the selection of five confirmed hits

(Figure 1B), that were designated ACP1 to 5. ACP1, ACP2, and

ACP3 were hits from the Maybridge library, while ACP4 and

ACP5 were hits from the Chembridge library. Interestingly,

ACP4 and ACP5 were analogous molecules, with identical

structures that only differed in the modification of an aromatic

group (Figure 1B).

Characterization of ACP-Mediated ClpP ActivationTo assess and compare the potency of ACPs, dose-response

analyses were carried out following the degradation of casein-

FITC for 6 hours (Figure 2A). The results were evaluated using

a quantitative measure, which we developed and named the

relative degradation index (RD), defined as follows:

RD=

�D4ClpP+ compound

�after 6 hrs

��D4ClpP

�after 6 hrs�

D4E:coli ClpAP

�after 6 hrs

��D4E:coli ClpP

�after 6 hrs

: (1)

D4 is the change in fluorescence after 6 hr of starting the reac-

tion (see Experimental Procedures)measured using 485 nmexci-

tation and 535 nm emission, which primarily detects the signal

from casein-FITC. E. coli ClpAP was used as a benchmark for

maximum ClpP proteolytic activity. The ClpP in the numerator

can be from any other organism. Based on RD25measurements,

which refers to the measurement in the presence of 25 mM

compound, the ranking of the activators from strongest to weak-

est is as follows: ACP1, ACP5, ACP4, ACP2, and ACP3 (Fig-

ure 2B). It should be emphasized that the RD value reflects how


much casein is degraded after 6 hr and, in theory, need not

directly correlate with the Kd for the binding of a given compound

to casein or with degradation rates, although, in practice, we

observe a general correspondence between these parameters.

The activation of ClpP by the various ACPs was further

confirmed by direct observation of unlabeled casein degradation

on SDS-PAGE gels (see Figure S1A available online), which also

indicated that the rank order of compound-mediated ClpP

activation was consistent with the observations obtained from

fluorometric determinations.

In order to verify that the activators were acting on ClpP rather

than on the casein substrate directly, the ability of compound-

activated ClpP to degrade proteins from various organisms

was tested using a variety of model substrates and unrelated

proteins (Figures S1B and S1C). The stronger activators induced

ClpP to degrade a wider variety of protein substrates than the

weaker activators with no apparent specificity. However, the

proteins subject to extensive proteolytic cleavage, such as

Casein (Creamer et al., 1981), Tah1 (Zhao et al., 2005), CFTR

R-region (Ostedgaard et al., 2001), reduced carboxymethylated

a-lactalbumin (RCMLa), lN protein (Mogridge et al., 1998),

and a-synuclein (Tompa, 2009) are either considered to be

unstable or disordered proteins or to have disordered regions.

Conversely, well-folded substrates, such as GFP-SsrA and crea-

tine kinase, were degraded to a lesser extent or not at all, likely

due to the stability of their structures. Proteins that were only

clipped, but not completely degraded, by compound-activated

ClpP (a-M protein, ClpA, and ClpX) were all larger sized proteins.

Chemical Optimization of ACP1When tested for bactericidal properties against ten different

bacteria, several of the ACPs showed minimum bactericidal

concentrations (MBC in mg/ml) at relatively low concentrations

(Table 1). Although certain compounds were more effective

against some bacteria compared with others, it was generally

observed that gram-negative bacteria were more sensitive to

these compounds than gram-positives (Table 1).

To determine whether the effects observed were due to the

interaction of the ACPs with ClpP in the bacteria, we constructed

N. meningitidis H44/76 strain (Nme H44/76, a genetically trac-

table strain) in which the clpP gene was disrupted as described

in Experimental Procedures. The disruption of clpP was verified

by PCR (Figure S2A). The susceptibility of WT and DclpP Nme

H44/76 strains to the different compounds was assessed using

either the disk diffusion plate assay or the liquid culture MBC

assay. A differential effect was observed for ADEP1A and

ADEP1B (Figure S2B) in addition to ACP1 and, to a weaker

extent, ACP4 (not shown). For ACP1, the MBC for WT Nme

H44/76 was 128 mg/ml, while for DclpP it was 64 mg/ml. The

Nme H44/76 DclpP strain was more resistant to the effect of

the compounds compared with WT as would be expected

if the compounds are primarily targeting the bacterial ClpP.

These observations prompted us to implement medicinal

chemistry rationales aimed at the refinement and optimization

of primary hits.

The chemical structures of ACP1–5 (Figure 1B) do not show

any obvious structural similarities with each other, except for

ACP4 and 5, or with the ADEPs (Figure S2C). ACP4 and 5 were

considered to be unsuitable for further optimization as they do


Table 1. Minimum Bactericidal Concentration of Compounds

PMBN

ADEP1A ACP1 ACP2 ACP3 ACP4 ACP5 ACP1a ACP1b

+ � + � + � + � + � + � + � + �N. gonorrhoeae

(N.279)

0.125a 0.125a >256 >256 32b 32b >256 >256 4a 4a >256 >256 128 >256 >256 >256

N. meningitidis

(MC58)

0.25a 0.25a 64b 64b >256 64b >256 >256 32b 32b >256 >256 64b 64b 16b 16b

H. influenzae

(H2192)

8a 128 64b >256 8a >256 >256 >256 4a 128 32b >256 32b >256 8a >256

E. coli (DH5a) 16b >256 >256 >256 8a >256 >256 >256 16b >256 16b >256 >256 >256 >256 >256

S. typhimurium

(SL1344)

>256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256

P. aeruginosa

(PAO1)

16b >256 >256 >256 8a >256 >256 >256 >256 >256 >256 >256 >256 >256 >256 >256

S. aureus

(ATCC 29213)

N/A 4a N/A >256 N/A >256 N/A >256 N/A >256 N/A >256 N/A >256 N/A >256

S. pneumoniae

(ATCC 49619)

N/A 16b N/A >256 N/A 16b N/A >256 N/A 8a N/A >256 N/A N/A N/A N/A

L. monocytogenes

(EGD)

N/A 0.125a N/A >256 N/A 32b N/A >256 N/A >256 N/A >256 N/A N/A N/A N/A

M. smegmatis

(mc2155)

N/A >256 N/A >256 N/A >256 N/A >256 N/A 128 N/A 256 N/A >256 N/A >256

128 or more except where indicated with a footnote. The membrane permeabilizing agent polymyxin B nonapeptide, PMBN, was added at 120 mg/ml

where indicated.a 0.125–8.b 16–64.

Chemistry & Biology


not have drug-like structures and would be more challenging to

assess in a structure activity relationship (SAR) studies (Keller

et al., 2006). ACP2 would be relatively straightforward to eval-

uate in an SAR study, but the protected tripeptide derivative

framework was not considered optimal as a potential small-

molecule lead. Instead, ACP1 and 3 both display drug-like struc-

tures amenable to further optimization, synthesis, and SAR

studies. In light of the higher RD25 score for ACP1 (Figure 2B)

and the fact that it seems to indeed target ClpP in the cell as

described above, we decided to explore chemical diversity

around the ACP1 molecule.

ACP1 satisfies Lipinski’s rule of five (Lipinski et al., 2001) and

has a topological polar surface area of 75.6, calculated molar

refractivity of 10.4, and Clog P of 3.98. The structure of ACP1

consists of a central b-amido sulfone core appended with

a western electron deficient pyridyl ring and an eastern hydro-

phobic tail incorporating a phenylthioether group (Figures 1B

and 3A). The natural synthetic disconnection point chosen for

the synthesis of ACP1 analogswas the amide linkage (Figure 3A).

Analogs were synthesized using a late-stage amide-bond form-

ing reaction between the eastern amine and the western

b-sulfonyl carboxylic acid (Figure 3B) by applying standard

synthetic protocols (see Experimental Procedures) and, subse-

quently, evaluated by measuring their RD25 values. We used

the natural products ADEP1 factor A and B (Figure S2C) isolated

from Streptomyces hawaiiensis (see Experimental Procedures)

as a reference. ADEP1A is a better ClpP activator than ADEP1B

(Figure 3C).

Of more than 70 analogs generated, one compound, which we

termed ACP1b (Figure 1B), was found to exhibit an RD25 value

1170 Chemistry & Biology 18, 1167–1178, September 23, 2011 ª201

slightly higher than that of ADEP1A (Figure 3C). For comparison,

another analog, which we termed ACP1a (shown in Figure 1B),

displayed a lower RD25 value than that of ACP1b but compa-

rable to that of ADEP1B (Figure 3C). ACP1a incorporates a sulfur

to methylene substitution in the eastern tail, and ACP1b has an

ortho-chloro substituent in the arylthioether ring (Figure 1B).

The extent of casein degradation upon activation of ClpP by

these compounds is shown in Figure 3D and is generally consis-

tent with the RD25 results. It has previously been shown that

ADEPs activate E. coli ClpP to degrade casein with reduced

processivity (Brotz-Oesterhelt et al., 2005; Kirstein et al., 2009).

This was also true for ACP1, ACP1a, and ACP1b (Figure 3E).

The patterns of appearance and disappearance of the degrada-

tion intermediates suggests similarities between the general

mechanisms of activation by the different compounds.

Notably, among the different bacterial species tested, ACP1b

showed an MBC value against H. influenzae that was compa-

rable to that of ADEP1A and much lower than that of ACP1

and ACP1a (Table 1). ACP1b also had improved MBC value

against N. meningitidis MC58 compared with ACP1 and

ACP1a, but not ADEP1A (Table 1). The MBC values for ACP1b

were 32 and 64 mg/ml for WT and DclpP Nme H44/76 described

above, respectively. Hence, while the WT is more sensitive to

ACP1b than the DclpP strain indicating that ClpP is indeed

a target for the compound in the cell, ACP1b might have other

cellular targets and further optimization will have to concentrate

on increasing specificity.

Although ACP1b was optimized against E. coli ClpP,

E. coli DH5a was resistant to this compound (Table 1), which is

consistent with earlier results (Brotz-Oesterhelt et al., 2005).

1 Elsevier Ltd All rights reserved

C D

0.0

0.2

0.4

0.6

0.8

1.0

1.2

RD

25

ADEP1A

ADEP1BACP1

ACP2ACP3

ACP4ACP5

ACP1a

ACP1b

N

F3C

SO O O

OH

H2NX

Ar

N SO O O

NHX

ArPyBOP,

iPr2NEt, DMF

F3C

Ar1S

O O O

NHX

Ar2

amidedisconnection

westernfragment

easternfragment

A

B

Time (hours)0 1 2 3 4 5 6

ACP1b

ACP1a

ACP1

ADEP1A

Casein

ADEP1B

No compound

ClpAP

Time (min) 0 2 105 20 603002 105 20 6030ADEP1A ACP1

- casein- ClpP

MW

(kD

a)

45 35 25

-

18

-

14

-

--

-4

2 105 20 60300ACP1a

2 105 20 60300ACP1b

2 105 20 60300no compound

E

Figure 3. Chemical Optimization of ACP1

(A) General chemical structure of the ACP1 analogs.

(B) A schematic of the chemical reaction to synthesize the ACP1 analogs using PyBoP mediated amide bond formation between 2-methyl-2-((5-(trifluoromethyl)

pyridin-2-yl)thio)propanoic acid and primary amines.

(C) RD25 values of ADEP1A, ADEP1B, ACP1–5, ACP1a, and ACP1b (see also Figure S2). Error bars represent the standard deviations from the average of three

repeats.

(D) Shown is the degradation of unlabeled casein by compound-activated ClpP followed on SDS-PAGE gels.

(E) The formation of intermediates during casein degradation by compound-activated ClpP.

Intermediate species, indicated by the parenthesis, resulting from casein degradation were resolved on 18% SDS-PAGE gels.

Chemistry & Biology


Brotz-Oesterhelt et al. (2005) did not see an effect of ADEPs on

WT E. coli, but rather on an E. coli strain deleted of the multidrug

efflux pump AcrA in the presence of the outer membrane perme-

abilizer polymyxin B nonapeptide (PMBN). Hence, we combined


ACP1b with several other known drugs to enhance sensitivity.

Using this approach, ACP1b was found to affect cell growth of

E. coli MC4100 strain in the presence of 20 mM of the uncoupler

ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP)


A

-0.20

-0.15

-0.10

-0.05

0

0 1000 2000 3000 4000

-3.5-2.0-2.5-2.0-1.5-1.0-0.5

0

45000.05

Molar Ratio0 10 20 30 40 50 60 70

Time (s)

80 90

ClpP14

- ACP1b

Molar Ratio0 10 20 30 40 50 60 70

kcal

mol

e-1

-8

-6

-4

-2

0

Time (s)0 1000 2000 3000 4000

µcal

sec

-1

-0.6-0.5-0.4-0.3-0.2-0.10.00.1

ClpP14

- ADEP1A

-7-6-5-4-3-2-10

-0.4

-0.3

-0.2

-0.1

0.0

0.1

Time (s)0 1000 2000 3000 4000

Molar Ratio0 10 20 30 40 50 60 70

ClpP14

- ADEP1B

Compound Kd (µM) n ΔH (kcal mol-1) ΔS (cal mol-1 K-1)ADEP1A 0.3 (0.1) 14.6 (0.4) -7.7 (0.3) 4.8 (1.2)ADEP1B 0.7 (0.2) 11.0 (0.5) -6.6 (0.4) 6.8 (1.5)

3.2 (0.5) 16.1 (1) -4.4 (0.4) 10.7 (1.2)ACP1b

χ2/DoF160683854479150

0

5

10

15

20

0 5 10 15 20 25[ACP1b] (µM)

S0.5h = 4.28 (0.01)

= 3.1 (0.2) µM

0 5 10 15 20 25

case

in d

egra

datio

n ra

te[(r

elat

ive

fluor

esce

nce

chan

ge) µ

M-1

s-1

]

[ADEP1A] (µM)

0

5

10

15

20

h = 3.86 (0.02)= 2.4 (0.2) µMS0.5

B

Figure 4. Determination of Binding Affinity of Compounds to ClpP

(A) ITC binding curves for ClpP-ADEP1A, ADEP1B, or ACP1b interaction are shown. Results for the fit of the data to a one set of identical independent binding site

model is given in the table. Numbers in parentheses refer to standard deviations. c2/DoF refers to chi-square divided by the degrees of freedom and indicates the

quality of the fit (see also Figure S3).

(B) Cooperativity of binding of ADEP1A and ACP1b to ClpP was determined bymeasuring the change in casein degradation rate by compound-activated ClpP as

a function of compound concentration. Error bars represent the standard deviations from the average of three repeats.

Chemistry & Biology


(FigureS2D). This sensitivitywasnot observed forMC4100DclpP

strain consistent with ClpP being a target for ACP1B in the cell.

The above observations indicate that the antibacterial proper-

ties of ACP1 can be improved through relatively simple chemical

modifications and that further efforts will have to concentrate on

enhancing specificity.

Determination of Binding Affinity and Stoichiometryof ACP1b Interaction with ClpPThe dissociation constant (Kd) for the binding of ACP1b to ClpP

was measured using isothermal titration calorimetry (ITC) and

was found to be about 3.2 ± 0.5 mM, which is comparable to,

albeit slightly higher, than that of ADEP1A (0.3 ± 0.1 mM)

and ADEP1B (0.7 ± 0.2 mM) (Figure 4A). The ITC data were fit

by allowing the number of binding sites (n) to vary (Figure 4A),


by fixing n to a whole number (Figure S3A), or by fixing n to 14

(Figure S3B). The results suggest that there are 14 binding sites

for the ADEPs and ACP1b on ClpP14. In comparison, ACP1 has

a dissociation constant for ClpP of about 130 mM as determined

by Surface Plasmon Resonance (SPR) measurements (Fig-

ure S3C), whereas ITC was unsuitable to measure ACP1 affinity

for ClpP due to weak binding. The similar Kd values obtained for

the ADEPs by SPR (Figure S3C) and ITC (Figure 4A) clearly indi-

cate that ACP1 has lower affinity than the ADEPs for ClpP. SPR

experiments were also carried out for ACP2–5, and the Kd values

obtained were much weaker (>150 mM) than that of ACP1 as

expected (data not shown).

Measurement of casein degradation rate by compound-acti-

vated ClpP as a function of ADEP1A or ACP1b concentration re-

sulted in a sigmoidal saturation curve with S0.5 of 2–3 mM and


A

Nor

mal

ized

c(S

)

0 200 400 600Rel

ativ

e flu

ores

cenc

e ch

ange

Time (s)

ClpPΔ31

ClpPΔ31+ACP1

ClpPΔ31+ACP1b

ClpPΔ31+ACP1a

full lengthClpP

0

0.1

0.2

0.3

0.4

0.5

C

0

0.2

0.8

0.6

0.4

0 5 10 15 20

full

leng

th C

lpP

ClpP14

ClpP7

s20,w (S)

0.5

1.0

1.5

2.0DMSOACP1

Clp

PΔ31

ClpPΔ3114

ClpPΔ317

2

3

1

ACP1b

45

ACP1a

6 70

9

10

Peak 1 2 3 4 5 6 7

9 10

8

MW (kDa) 134 140 277 145 297 145 282

148 303

298

Frictional ratio 1.35780 1.38695 1.38695 1.42471 1.42471 1.34795 1.34795

1.33438 1.33438

1.35871

s (S)

4.322 4.3536.860 4.337 7.000 4.391 7.250

4.696 7.587

7.220

7.307 6.994 10.943 6.918 11.167 7.004 11.564

7.491 12.103

11.517

ADEP1A8

60

-6

0.5

5.9 6.00

6.1no compound

Abso

rban

ce @

280

nm

Res

idua

ls

Radius (cm)

Kd = 8.59x10-4 M

0.5

5.9 6.00

6.1

420

-2-4

ACP1aAb

sorb

ance

@ 2

80 n

mR

esid

uals

Radius (cm)

Kd = 3.68x10-5 M

420

-2

0.5

5.9 6.0 6.1

ACP1b

Radius (cm)

Abso

rban

ce @

280

nm

Res

idua

ls

Kd = 3.77x10-6 M

0

B

ADEP1A

s20,w (S)

[(16.7-4.93)x10-4 M]

[(4.58-2.98)x10-5 M]

[(4.65-3.04)x10-6 M]

Figure 5. Effect of ACP on ClpP Oligomeric Stability

(A) Sedimentation velocity analytical ultracentrifugation of full length ClpP (46 mM) andClpPD31 (51 mM) at 4�C shown using the continuous distributionmodel c(S)

versus s20,w, scaled to initial absorbance. ACP1, ACP1a, ACP1b, and ADEP1A (at 100 mM) promote the tetradecamerization of ClpPD31 to different extents. The

table lists the sedimentation coefficients, frictional ratios, and molecular weights corresponding to the various peaks.

(B) Sedimentation equilibrium profiles and the corresponding distribution of residuals for 51 mMClpPD31 in the absence of compound (top), or in the presence of

100 mM ACP1a (middle) or 100 mM ACP1b (bottom). The solid lines represent the best fit to a monomer-dimer model, with the heptameric ClpP (MW of 138,544

Da) treated as the monomer. The resulting dissociation constants are given for each data set. The numbers in brackets give the range of Kd values for the 95%

confidence interval.

(C) The effect of compounds (at 100 mM) on the peptidase activity of ClpPD31 (1 mM) is shown. The defective peptidase activity of thismutant is partially recovered

by the presence of the ACP compounds.

Chemistry & Biology


a Hill coefficient, h, of about 4 (Figure 4B). The data suggest that

these compounds act on ClpP in a similar manner and bind to the

protease cooperatively or promote cooperative allosteric transi-

tions within the protease structure.

The ClpP Tetradecamer Is Stabilized by ACP BindingThermal melt experiments indicated that the ACP compounds

enhanced ClpP stability (not shown). Hence, we asked whether

the compounds affect ring-ring interactions within the ClpP


double-ring tetradecamer. An N-terminal truncation mutant of

E. coli ClpP was originally constructed to open the central pore

of the protease in an attempt to mimic a proposed mechanism

of compound activation. However, this mutant, ClpPD31, mainly

eluted as a heptameric single ring upon size exclusion chro-

matography and had no significant peptidase activity. In order

to verify the oligomeric state of this protein, we employed sedi-

mentation velocity analytical ultracentrifugation (Figure 5A). The

results indicate that the ClpPD31 mutant has a sedimentation


Chemistry & Biology


coefficient corresponding to that of a heptamer; by comparison,

WT ClpP has a sedimentation coefficient corresponding to a tet-

radecamer with a small proportion of heptamers (Figure 5A).

Addition of the different compounds shifts the oligomeric state

of ClpPD31 to that of a tetradecamer (Figure 5A). The proportion

of ClpPD31 tetradecamers formed correlates with the binding

affinity of the compounds. For example, ACP1b led to the forma-

tion of a higher proportion of tetradecamers than ACP1, while

ADEP1A led to the formation of the ClpPD31 tetradecamer

exclusively. It should be noted that integration of the normalized

c(S) signal over the s-values, which cover the range of sediment-

ingmaterial, shows that the total ‘‘loading signal’’ is similar for the

different curves with an average total loading signal of 0.52

(±0.12). The differences in the normalized c(S) peak heights is

mainly because some peaks are broader than others for the

various treatments of ClpPD31. Slight shifts in alignment of the

ClpPD31 tetradecamer sedimentation coefficient in the pres-

ence of the various compounds might indicate some disas-

sembly of the tetradecamer under the sedimentation velocity

nonequilibrium conditions or may be attributed to different

hydrodynamic properties of the tetradecamers formed. We

further verified the oligomeric state of the ClpPD31 in the pres-

ence of ACP1a or ACP1b using sedimentation equilibrium exper-

iments (Figure 5B). Fits of the data to monomer-dimer equilib-

rium, assuming the monomer is heptameric ClpPD31 with a

molecular weight of 138,544 Da, clearly indicate that the pres-

ence of ACP1a or ACP1b decreases the dissociation constant

of the ClpP tetradecamer with ACP1a having a lower Kd than

ACP1b, which, in turn, is lower than that in the absence of the

compound (Figure 5B).

Without the addition of compound, ClpPD31 had little

protease activity (Figure 5C). Addition of ACP1, ACP1a,

ACP1b, or ADEP1A promoted the formation of tetradecameric

ClpPD31 (Figures 5A and 5B) and resulted in a catalytically active

ClpPD31 (Figure 5C). These results strongly suggest that the

ACP compounds stabilize ClpP by promoting the formation of

the double-ring structure.

ACP Binding Sites on ClpPThe ADEPs have been found to bind in a hydrophobic pocket (H

pocket) on the ClpP apical surface in which the IGF loop of the

ClpX/ClpA ATPases also binds (Lee et al., 2010; Li et al.,

2010). As a result, the presence of the ADEPs inhibited or

reduced the ClpXP/ClpAP-mediated degradation of GFP-ssrA

by interfering with the binding of ClpX/ClpA to ClpP. As shown

in Figure 6A and Figure S4A, a similar effect was also seen for

the ACPs suggesting that these compounds either bind to or

allosterically modulate the H pocket of ClpP.

Two binding pockets on ClpP of equal probability were pre-

dicted by computational procedures implemented in DOCK6.3

software (Lang et al., 2009) (see Experimental Procedures; Fig-

ure 6B; Table S1), namely, the H pocket and a separate pocket,

which we have named the C pocket, featuring a larger number of

charged residues. The H pocket is composed of residues V42,

F44, L62, Y74, Y76, I104, F126, L203, and R206 from one

subunit, and L37 and F96 from the neighboring subunit (using

Swiss-Prot numbering). The C pocket comprises residues Y90,

M94, Q95, D100, V101, and H170 of one subunit and residues

H205 and N207 of the neighboring subunit. Interestingly, the H


and C pockets are separated by residues at the very C terminus

of ClpP, corresponding to amino acids 203–207. While ADEPs

cocrystallized with ClpP were found to be bound to the H pocket

(Lee et al., 2010; Li et al., 2010), all the ACP compounds

were docked well to both the H and C pockets (Figure 6B), and

their docking scores differed little between the two pockets

(Table S1).

To further validate the results of the docking analyses, we

generated and tested mutants of the H pocket (Y76W [Lee

et al., 2010], F126A [Bewley et al., 2006], and L203E), C pocket

(Q95A, D100A, and H170A), as well as both the H and C pockets

(Figure 6B). Peptidase assays were first performed to assess

whether the mutations affected ClpP active sites. Except for

Y76W, H170A, and Q95A/F126A, the mutants exhibited pepti-

dase activity similar to that of the wild-type (WT) protein (Fig-

ure S4B). Intriguingly, we observed that the effects of stronger

ClpP activators, such as ADEP1A and ACP1b, markedly de-

creased in the presence of mutations confined to the H pocket,

whereas weaker activators appeared to be affected by muta-

tions in both H and C pockets (Figure 6B). These findings, taken

together, suggest that a successful strategy aimed at developing

stronger ClpP activators is that of developing small molecules

interacting with ClpP within the H pocket.

DISCUSSION

In our study, we have successfully identified five new com-

pounds representing four different structural classes (Figure 1B)

that activate ClpP protease and have bactericidal properties.

With the exception of ACP4 and ACP5, these compounds have

no apparent structural similarities to each other or to the previ-

ously reported ADEPs (Brotz-Oesterhelt et al., 2005), dramati-

cally increasing the repertoire of compounds activating this

protease. The optimization of ACP1 resulted in compounds

that had in vitro ClpP activation properties close to that of the

natural product ADEPs (Figure 3C). Importantly, these optimized

compounds have good bactericidal properties (Table 1), sug-

gesting potential applications as therapeutics. In particular, our

data indicate that these compounds may be dramatically

improved through a much tighter affinity for ClpP and specific

targeting within the ClpP H pocket.

Based on the results of Figures 4–6, we propose that the iden-

tified ACP compounds share a similar mechanism of ClpP acti-

vation as the ADEPs. They all stabilize ClpP and promote the

formation of the double-ring structure, albeit through a yet

unknownmechanism. The apparent lack of substrate preference

(Figures S1B and S1C) that the compounds confer on ClpP in

addition to the similarities in degradation patterns (Figure 3E) is

also indicative of similarities in mechanism of ClpP activation.

The compounds prevent the binding of ClpP to its associated

unfoldase (Figure 6A; Figure S4A) and, at the same time, activate

nonspecific proteolysis probably through the opening of the

axial pores.

The recently solved cocrystal structures of ADEP bound to

B. subtilis and E. coli ClpP (Lee et al., 2010; Li et al., 2010)

show ADEP binding to the H pocket of the protease. Our results

are in general agreement with these findings; however, we

propose the presence of an additional pocket, the C pocket,

that is more charged (Figure 6B) and, as suggested by our


Figure 6. ACP Binding Sites

(A) Shown is the inhibition of ClpXP-mediated GFP-ssrA degradation by ACPs and ADEPs added at 100 mM monitored on SDS-PAGE gels. ClpP was

preincubated with compound before the addition of ClpX (see also Figure S4A).

(B) Surfacemodel of ClpP is shown on the top left. Four neighboring subunits are colored in alternating blue and green. The H pockets are colored in purple, while

the C pockets are colored in yellow. The bottom left panel shows a close up view of the predicted compound binding conformations in the two ClpP pockets. The

five ACP compounds are overlaid in the binding pockets (see also Table S1). ClpP is shown as a surfacemodel and the compounds are shown as stick models. C,

N, O, S, F, Cl, Br, and H in the compounds are colored in gray, blue, red, yellow, cyan, green, purple, and white, respectively. The ClpP surface is colored

according to the electrostatic surface potential and ranges from red (potential of �4kT) to blue (potential of +4kT) calculated using DelPhi (Rocchia et al., 2002).

Also shown on the right panels are stickmodels of ACP1 docked into the H and C pockets of ClpP drawn as ribbons colored by chain. For ACP1, C, N, O, S, F, and

H are colored in orange, blue, red, yellow, cyan, and white, respectively. All molecular graphics figures were prepared using the program PyMOL.

(C) Effect of mutations in the H and C pockets on ClpP activation by compounds measured using RD25. Data shown represent the average of three repeats and

error bars give the standard deviations on the measurements (see also Figure S4B).

Chemistry & Biology


Chemistry & Biology 18, 1167–1178, September 23, 2011 ª2011 Elsevier Ltd All rights reserved 1175

Chemistry & Biology


mutational analysis, involved in compound binding (Figure 6C). It

is likely that the binding of compounds in the H pocket mimics

the binding of ATPase chaperones and causes conformational

changes in the ClpP axial pore, which normally occur upon

binding of the Clp ATPase. The ADEP compounds, being much

larger and bulkier, make more contacts in the ClpP binding

pocket than the ACPs. The ACP compounds make fewer

contacts, yet still induce ClpP activation. How these dissimilar

compounds can cause the same effects warrants further inves-

tigation. Cocrystallization trials are currently under way to gain

further understanding of the interaction of these compounds

with ClpP.

In summary, our study identified new activators of ClpP that

we term ACPs. These activators can be purchased relatively

cheaply from regular vendors allowing any group to study this

phenomenon without the need to rely on difficult purification

procedures of natural products or complicated chemical

synthesis methods. We expect that activators of self-compart-

mentalizing proteases will be important players in future drug

development efforts.

SIGNIFICANCE

With the rise of antibiotic-resistant bacteria, there is an

urgent need for the development of new compounds having

novel mode of function. The ability to dysregulate ClpP

serine protease activity represents a novel approach for

the development of new antibacterials. ClpP activity is

tightly regulated by bound ATPase chaperones. These

ATPases typically form hexameric rings that bind on one or

both ends of the ClpP double-ring cylinder. The ATPases

select target substrates, which are then unfolded and

threaded through the ATPase ring and into the ClpP cylinder

for degradation. Threading into the ClpP proteolytic

chamber occurs through narrow axial pores that do not

allow structured proteins to pass, hence, the requirement

for protein unfolding. Acyldepsipeptides, ADEPs, were

recently found to bind ClpP and open the axial pores of the

protease allowing ClpP to degrade folded proteins indepen-

dent of its chaperone resulting in unregulated degradation

of protein substrates. ADEPs were found to have antibacte-

rial activity and were originally purified from Streptomyces.

However, the chemical synthesis of these compounds is

quite challenging. In order to identify novel compounds

that activate ClpP, we used a high-throughput screening

approach with a fluorescence-based readout. The assay

employed fluorescein isothiocyanate-labeled casein as the

proteolytic target of the Escherichia coli ClpP protease.

Five structurally diverse compounds were identified to acti-

vate ClpP that we named Activators of Self-Compartmental-

izing Proteases 1 to 5 (ACP1–5). Their chemical structures

differ significantly from the structures of the ADEPs. The

chemical optimization of ACP1 resulted in analogs having

improved bioactivity and bactericidal effects. The com-

pounds were found to stabilize the ClpP double-ring struc-

ture and to bind in pockets on the ClpP apical surface.

Hence, our study provides the basis for the development

of novel antibiotics based on the activation and dysregula-

tion of ClpP activity using different structural scaffolds.


EXPERIMENTAL PROCEDURES

Protein Expression and Purification

All proteins were expressed from IPTG inducible promoters. ClpP constructs

were expressed in BL21(DE3)1146D strain, which lacks the gene for chromo-

somal E. coli ClpP. All other constructs were expressed in BL21(DE3)Gold

(Stratagene). Untagged WT and mutant E. coli ClpP, ClpX, and GFP-SsrA

were expressed and purified as previously described (Wojtyra et al., 2003).

ClpA was expressed and purified as described in Lo et al. (2001). All His-

tagged proteins were purified on Ni-NTA agarose resin (QIAGEN) according

to the manufacturer’s protocols. If possible, the tag was removed using the

tobacco etch virus (TEV) protease. Protein concentrations were determined

by absorbance at 280 nm with extinction coefficients calculated using

ProtParam (http://ca.expasy.org/tools/protparam.html).

Tah1 (Zhao et al., 2008) and lO (Wojtyra et al., 2003) were purified as

previously described. pHF010 plasmid encoding H6-lN was a kind gift from

Dr. Irene Lee (Case Western Reserve University) and the protein was purified

according to published protocols (Patterson-Ward et al., 2009). CFTR

R-domain and a-synuclein were gifts from Dr. Julie Forman-Kay (University

of Toronto). The Neisseria a-M protein was from Mr. Shekeb Khan from

EFP’s group. Reduced carboxymethylated a-lactalbumin was a gift from

Dr. John Glover (University of Toronto). Creatine kinase, a-casein, casein

fluorescein isothiocyanate (casein-FITC, type II, 20–50 mg FITC/mg, catalog

number C3777), and N-Succinyl-Leu-Tyr-7-amido-4-methylcoumarin (Suc-

LY-AMC) were purchased from Sigma-Aldrich.

Chemical Libraries and High-Throughput Screening

The libraries employed for the screening campaigns were composed of exper-

imental bioactives, pharmacologically active chemicals and natural products,

off-patent marketed drugs, and small molecules with drug-like properties.

Samples were obtained from the following, commercially available collections:

LOPAC 1280 (Sigma, 1280 samples), Prestwick Chemical library (Prestwick

Chemical, France, 1120 samples), SPECTRUM collection (MicroSource,

2000 samples), Maybridge Screening collection (Maybridge-Thermo Fisher

Scientific, UK, 50,000 samples), and Chembridge DIVERSet (ChemBridge

Corp, 10,000 samples). In all instances, samples stored in 384-well plates as

1 or 5 mM solutions in 100% DMSOwere transferred to assay plates in a fixed

volume of 200 nl by a pin-tool (V&P Scientific). Screens were conducted using

a fully automated procedure run on a DIM4 flipmover platform (Thermo Elec-

tron Corp) equipped with a Biomek FX liquid handler (Beckman, USA) and

a PHERAStar detection system (BMG Labtech, Germany). The reaction for

the screening assay contained 20 mM compound, 3.6 mM ClpP, and 4.5 mM

casein-FITC in buffer A (25 mM Tris HCl [pH 7.5], and 100 mM KCl) at 37�C.Compounds thatwere identified andconfirmed tobe hitswere obtained from

the following companies: ACP1, ACP2, and ACP3 are from Maybridge (May-

bridge, Thermo-Fisher Scientific, catalog number BTB09142, JFD02943, and

KM11066, respectively); ACP4 and ACP5 are from Chembridge (ChemBridge

Corporation, catalog number 5107477 and 5107473, respectively).

Activity Assays

To measure the RD index, each reaction consisted of 3.6 mMClpP and a spec-

ified amount of compound in buffer A. The reactions were preincubated for

10 min at 37�C before 4.5 mM casein-FITC and 15.5 mM unlabeled casein

were added. ClpAP-dependent degradation of the same casein-FITC

substrate was used as a control. Each ClpAP reaction contained 3.6 mM

ClpP, 3 mM ClpA, and 0.3 mM ATP in buffer B (25 mM HEPES [pH 7.5],

20 mM MgCl2, 30 mM KCl, 0.03% Tween 20, and 10% glycerol), and an

ATP-regenerating system (13 units/ml of creatine kinase and 16 mM creatine

phosphate). The reaction was started by adding casein-FITC. Reactions

were incubated at 37�Cand the fluorescence (485 nmexcitation, 535 nmemis-

sion) was monitored every 15 min for 6 hr on a PHERAStar detection system.

Casein-FITC degradation by compound-activated ClpP after 6 hr at 37�C was

compared with the ClpAP-dependent casein-FITC degradation after 6 hr at

37�C, using Equation 1.

The effects of the compounds on the degradation of GFP-ssrA by ClpXP

were also analyzed on SDS-PAGE gels. Each reaction contained 1.2 mM

ClpP, 3.9 mM GFP-ssrA, 3 mM ATP, and 100 mM compound in buffer C

(25 mM HEPES [pH 7.5], 5 mM MgCl2, 5 mM KCl, 0.03% Tween 20, and


http://ca.expasy.org/tools/protparam.html

Chemistry & Biology


10% glycerol), and an ATP-regenerating system. The reaction mixtures were

preincubated for 3 min at 37�C before 1 mM ClpX was added. Samples were

taken at various time points, stopped by boiling in 2% SDS, and resolved on

12% SDS-PAGE gels.

Peptidase activity of ClpP was measured by the ability of ClpP to cleave the

dipeptide Suc-LY-AMC. Each reaction contained 1 mMClpP in buffer D (50mM

Tris HCl [pH 8], 200 mM KCl, and 1 mM DTT). ClpP was incubated for 3 min at

37�Cbefore Suc-LY-AMCwas added to a final concentration 0.5mM. Fluores-

cence (350 nmexcitation, 460 nm emission) of the released AMCwas detected

on the PHERAStar system.

For the Hill plot analysis of Figure 4B, each reaction consisted of 3.6 mMClpP

and specified amount of compound in buffer A. The reactions were pre-incu-

bated for 10 min at 37�C before 4.5 mM casein-FITC was added. Reactions

were incubated at 37�Cand the fluorescence (485 nmexcitation, 535 nmemis-

sion) was monitored every 3 s for 5 min on an EnSpire Multilabel Plate Reader

(PerkinElmer).

N. meningitidis H44/76 clpP Insertional Mutagenesis

To construct the N. meningitidis H44/76 clpP insertional mutant vector, desig-

nated Nme-clpP-ery, primers f2 and r2 shown in Figure S2A were used to

amplify a fragment of the genome that contained the clpP gene and the result-

ing PCR fragment was cloned into the pTrc99A vector. Using the f3 and r1

primers (Figure S2A), an erythromycin cassette with EagI ends was amplified

from the pFLOB4300 vector (Litt et al., 2000) and the resulting PCR fragment

was cloned into the EagI restriction site located at the 50 end of the clpP gene

within the Nme-clpP vector. The N. meningitidis H44/76 clpP insertional

mutant was generated by transforming WT N. meningitidis H44/76 with the

Nme-clpP-ery vector using an electroporation protocol adapted from (Dillard,

2006). PCR verification of the mutant strain is shown in Figure S2A.

Determination of Minimum Bactericidal Concentrations

Bacterial strains used included both gram-negative and gram-positive

bacteria: Escherichia coli DH5a, Salmonella typhimurium SL1344,

Pseudomonas aeruginosa PAO1, Haemophilus influenzae H2192, Neisseria

gonorrhoeae N.279, Neisseria meningitidis H44/76, Neisseria meningitidis

MC58, Staphylococcus aureus ATCC 29213, Streptococcus pneumoniae

ATCC 49619, and Listeria monocytogenes EGD. Due to the lowwater solubility

of the ADEP and ACP compounds, minimum bactericidal concentrations

(MBC) values were determined by plating compound-treated bacteria on

agar plates without compound. All compounds were diluted in Brain Heart

Infusion (BHI) medium with 1% Isovitale X (Becton Dickinson). All bacteria

were also inoculated into BHI. A 2-fold dilution series of each compound

was created in 96-well plates, with and without 120 mg/ml of the membrane

permeabilizing agent polymyxin B nonapeptide, PMBN (Sigma-Aldrich) (Ofek

et al., 1994; Tsubery et al., 2002). All bacterial suspensions were pelleted,

resuspended in BHI, and added to the compound containing media.

H. influenzae, N. gonorrhoeae, and N. meningitidis were incubated 18–20 hr,

while the remaining strains were incubated 12 to 16 hr. H. influenzae,

N. gonorrhoeae, N. meningitidis, and S. pneumoniae were incubated in the

presence of 5% CO2. 2 ml of culture from the incubations were then plated

onto compound-free agar plates (H. influenzae, N. gonorrhoeae,

N. meningitidis for 18–20 hr; remaining strains for 12–16 hr) to determine the

bactericidal activity. H. influenzae was grown on chocolate agar;

N. gonorrhoeae on GC agar base; N. meningitidis, S. aureus, and

L. monocytogenes on BHI agar; S. pneumoniae on 5% sheep blood agar;

and E. coli, S. typhimurium, and P. aeruginosa on LB agar. The lowest concen-

tration of compound at which no bacterial growth was seen was designated as

the minimum bactericidal concentration.

Antimycobacterial activities of the ClpP activating compounds were investi-

gated using the 96-well plate broth microdilution method modified from

Wallace et al. (1986), followed by plating for viable bacteria. Compounds

were dissolved in 100% DMSO. M. smegmatis mc2155 cultures were grown

to an OD600 of 1.0–1.3 in Middlebrook 7H9 broth (Difco, BD Biosciences)

supplemented with 2% oleic acid, albumin, dextrose, and catalase (OADC

enrichment, BD Biosciences) as well as 0.2% glycerol and 0.5% Tween 80

to avoid clumping of bacteria. Cultures were then diluted 10-fold and 5 ml

were used to inoculate 100 ml of a 2-fold dilution series of compounds in the

range of 1 mg/ml to 256 mg/ml in 7H9 broth supplemented with OADC and


0.2% glycerol in the presence or absence of 12.5 mg/ml polymyxin B. Poly-

myxin B is known to enhance mycobacterial permeability to hydrophobic

compounds (Korycka-Machala et al., 2001). The solvent control, DMSO at

2% or less, showed no inhibitory effects on M. smegmatis growth. Plates

were incubated in a CO2 incubator for 8 days. Following incubation, dilutions

of sample aliquots were spread onMiddlebrook 7H11 plates (Difco, BD Biosci-

ences) supplemented with 2% OADC and 0.5% glycerol for determination of

bacterial viability.

Monitoring of E. coli Growth in the Presence of ACP Compounds

The E. coli strains used for growth curves shown in Figure S2D, MC4100 and

MC4100 DclpP::cat, were grown overnight in LB and then diluted into fresh LB

to OD600 of 0.02. One hundred microliters of the diluted cell culture was added

to 100 ml of LB containing ACP1b and CCCP (carbonyl cyanide 3-chlorophe-

nylhydrazone) (Sigma-Aldrich). Final concentrations were 128 mg/ml ACP1b

and/or 20 mM CCCP. The solvent control cultures contained 20 mM CCCP

and 2.2% DMSO in place of ACP1b. Growth was monitored overnight at

30�C in a Bioscreen-C incubator system (Growth Curves, USA).

Other Procedures

Additional methods are given in the Supplemental Experimental Procedures,

namely, subcloning and mutagenesis, surface plasmon resonance measure-

ments, isothermal titration calorimetry measurements, analytical ultracentrifu-

gation measurements, thermal denaturation of ClpP, isolation of ADEP1A and

ADEP1B from Streptomyces hawaiiensis NRRL 15010, docking procedure,

and chemical synthesis of ACP1 and analogs.

SUPPLEMENTAL INFORMATION

Supplemental Information includes four figures, one table, and Supplemental

Experimental Procedures and can be found with this article online at

doi:10.1016/j.chembiol.2011.07.023.

ACKNOWLEDGMENTS

This work was supported by the Canadian Institutes of Health Research

Emerging Team Grants from the Institute of Infection and Immunity (XNE-

86945) to A.D., E.F.P., J.L., S.G.-O., R.A.B., andW.A.H. M.E.B. is the recipient

of a fellowship from the Canadian Institutes of Health Research (CIHR) Stra-

tegic Training Program in Protein Folding and Interaction Dynamics: Principles

and Diseases. S.J.W. is Canada Research Chair Tier 1, funded by the Canada

Institute of Health Research.

Received: February 16, 2011

Revised: June 28, 2011

Accepted: July 13, 2011

Published: September 22, 2011

REFERENCES

Bewley, M.C., Graziano, V., Griffin, K., and Flanagan, J.M. (2006). The asym-

metry in the mature amino-terminus of ClpP facilitates a local symmetry match

in ClpAP and ClpXP complexes. J. Struct. Biol. 153, 113–128.

Bewley, M.C., Graziano, V., Griffin, K., and Flanagan, J.M. (2009). Turned on

for degradation: ATPase-independent degradation by ClpP. J. Struct. Biol.

165, 118–125.

Brideau, C., Gunter, B., Pikounis, B., and Liaw, A. (2003). Improved statistical

methods for hit selection in high-throughput screening. J. Biomol. Screen. 8,

634–647.

Brotz-Oesterhelt, H., Beyer, D., Kroll, H.-P., Endermann, R., Ladel, C.,

Schroeder, W., Hinzen, B., Raddatz, S., Paulsen, H., Henninger, K., et al.

(2005). Dysregulation of bacterial proteolytic machinery by a new class of

antibiotics. Nat. Med. 11, 1082–1087.

Creamer, L.K., Richardson, T., and Parry, D.A.D. (1981). Secondary structure

of bovine alpha s1- and beta-casein in solution. Arch. Biochem. Biophys. 211,

689–696.


http://dx.doi.org/doi:10.1016/j.chembiol.2011.07.023

Chemistry & Biology


Dillard, J.P. (2006). Genetic manipulation of Neisseria gonorrhoeae. Curr.

Protoc. Microbiol. Chapter 4, Unit 4A 2.

Gottesman, S., Maurizi, M.R., and Wickner, S. (1997). Regulatory subunits of

energy-dependent proteases. Cell 91, 435–438.

Hinzen, B., Raddatz, S., Paulsen, H., Lampe, T., Schumacher, A., Habich, D.,

Hellwig, V., Benet-Buchholz, J., Endermann, R., Labischinski, H., and Brotz-

Oesterhelt, H. (2006). Medicinal chemistry optimization of acyldepsipeptides

of the enopeptin class antibiotics. ChemMedChem 1, 689–693.

Jennings, L.D., Lun, D.S., Medard, M., and Licht, S. (2008). ClpP hydrolyzes

a protein substrate processively in the absence of the ClpA ATPase: mecha-

nistic studies of ATP-independent proteolysis. Biochemistry 47, 11536–11546.

Katayama, Y., Gottesman, S., Pumphrey, J., Rudikoff, S., Clark, W.P., and

Maurizi, M.R. (1988). The two-component, ATP-dependent Clp protease of

Escherichia coli. Purification, cloning, and mutational analysis of the ATP-

binding component. J. Biol. Chem. 263, 15226–15236.

Keller, T.H., Pichota, A., and Yin, Z. (2006). A practical view of ‘‘druggability’’.

Curr. Opin. Chem. Biol. 10, 357–361.

Kirstein, J., Hoffmann, A., Lilie, H., Schmidt, R., Rubsamen-Waigmann, H.,

Brotz-Oesterhelt, H., Mogk, A., and Turgay, A. (2009). The antibiotic ADEP re-

programmes ClpP, switching it from a regulated to an uncontrolled protease.

EMBO Mol. Med. 1, 37–49.

Korycka-Machala,M., Zio1kowski, A., Rumijowska-Galewicz, A., Lisowska, K.,

and Sedlaczek, L. (2001). Polycations increase the permeability of Myco-

bacterium vaccae cell envelopes to hydrophobic compounds. Microbiology

147, 2769–2781.

Lang, P.T., Brozell, S.R., Mukherjee, S., Pettersen, E.F., Meng, E.C., Thomas,

V., Rizzo, R.C., Case, D.A., James, T.L., and Kuntz, I.D. (2009). DOCK 6:

combining techniques to model RNA-small molecule complexes. RNA 15,

1219–1230.

Lee, B.G., Park, E.Y., Lee, K.E., Jeon, H., Sung, K.H., Paulsen, H., Rubsamen-

Schaeff, H., Brotz-Oesterhelt, H., and Song, H.K. (2010). Structures of ClpP in

complex with acyldepsipeptide antibiotics reveal its activation mechanism.

Nat. Struct. Mol. Biol. 17, 471–478.

Li, D.H.,Chung,Y.S.,Gloyd,M., Joseph, E.,Ghirlando,R.,Wright,G.D.,Cheng,

Y.Q., Maurizi, M.R., Guarne, A., and Ortega, J. (2010). Acyldepsipeptide

antibiotics induce the formation of a structured axial channel in ClpP: A model

for the ClpX/ClpA-bound state of ClpP. Chem. Biol. 17, 959–969.

Lipinski, C.A., Lombardo, F., Dominy, B.W., and Feeney, P.J. (2001).

Experimental and computational approaches to estimate solubility and perme-

ability in drug discovery and development settings. Adv. Drug Deliv. Rev. 46,

3–26.

Litt, D.J., Palmer, H.M., and Borriello, S.P. (2000). Neisseria meningitidis ex-

pressing transferrin binding proteins of Actinobacillus pleuropneumoniae

can utilize porcine transferrin for growth. Infect. Immun. 68, 550–557.

Lo, J.H., Baker, T.A., and Sauer, R.T. (2001). Characterization of the N-terminal

repeat domain of Escherichia coli ClpA-A class I Clp/HSP100 ATPase. Protein

Sci. 10, 551–559.


Maurizi, M.R., and Xia, D. (2004). Protein binding and disruption by Clp/

Hsp100 chaperones. Structure 12, 175–183.

Michel, K.H., and Kastner, R.E. (1985). A54556 antibiotics and process for

production thereof. (US Patent 4492650).

Mogridge, J., Legault, P., Li, J., Van Oene, M.D., Kay, L.E., and Greenblatt, J.

(1998). Independent ligand-induced folding of the RNA-binding domain and

two functionally distinct antitermination regions in the phage lambda N protein.

Mol. Cell 1, 265–275.

Ofek, I., Cohen, S., Rahmani, R., Kabha, K., Tamarkin, D., Herzig, Y., and

Rubinstein, E. (1994). Antibacterial synergism of polymyxin B nonapeptide

and hydrophobic antibiotics in experimental gram-negative infections in

mice. Antimicrob. Agents Chemother. 38, 374–377.

Ostedgaard, L.S., Baldursson, O., and Welsh, M.J. (2001). Regulation of the

cystic fibrosis transmembrane conductance regulator Cl- channel by its R

domain. J. Biol. Chem. 276, 7689–7692.

Patterson-Ward, J., Tedesco, J., Hudak, J., Fishovitz, J., Becker, J., Frase, H.,

McNamara, K., and Lee, I. (2009). Utilization of synthetic peptides to evaluate

the importance of substrate interaction at the proteolytic site of Escherichia

coli Lon protease. Biochim. Biophys. Acta 1794, 1355–1363.

Rocchia, W., Sridharan, S., Nicholls, A., Alexov, E., Chiabrera, A., and Honig,

B. (2002). Rapid grid-based construction of the molecular surface and the use

of induced surface charge to calculate reaction field energies: applications to

themolecular systems and geometric objects. J. Comput. Chem. 23, 128–137.

Tompa, P. (2009). Structural disorder in amyloid fibrils: its implication in

dynamic interactions of proteins. FEBS J. 276, 5406–5415.

Tsubery, H., Ofek, I., Cohen, S., Eisenstein, M., and Fridkin, M. (2002).

Modulation of the hydrophobic domain of polymyxin B nonapeptide: effect

on outer-membrane permeabilization and lipopolysaccharide neutralization.

Mol. Pharmacol. 62, 1036–1042.

Wallace, R.J., Jr., Nash, D.R., Steele, L.C., and Steingrube, V. (1986).

Susceptibility testing of slowly growing mycobacteria by a microdilution MIC

method with 7H9 broth. J. Clin. Microbiol. 24, 976–981.

Wojtyra, U.A., Thibault, G., Tuite, A., and Houry, W.A. (2003). The N-terminal

zinc binding domain of ClpX is a dimerization domain that modulates the chap-

erone function. J. Biol. Chem. 278, 48981–48990.

Zhang, J.-H., Chung, T.D.Y., and Oldenburg, K.R. (1999). A Simple Statistical

Parameter for Use in Evaluation and Validation of High Throughput Screening

Assays. J. Biomol. Screen. 4, 67–73.

Zhao, R., Davey, M., Hsu, Y.C., Kaplanek, P., Tong, A., Parsons, A.B., Krogan,

N., Cagney, G., Mai, D., Greenblatt, J., et al. (2005). Navigating the chaperone

network: an integrative map of physical and genetic interactions mediated by

the hsp90 chaperone. Cell 120, 715–727.

Zhao, R., Kakihara, Y., Gribun, A., Huen, J., Yang, G., Khanna, M., Costanzo,

M., Brost, R.L., Boone, C., Hughes, T.R., et al. (2008). Molecular chaperone

Hsp90 stabilizes Pih1/Nop17 to maintain R2TP complex activity that regulates

snoRNA accumulation. J. Cell Biol. 180, 563–578.


Documents

Activators of Cylindrical Proteases as Antimicrobials: Identification and Development of Small Molecule Activators of ClpP Protease